Rolph, S; Mondain-Monval, T.O.; August, T.; Jarvis, S.G.; Wright, E.; Fox, R.; Pocock, M.J.O.
Species richness and recording priority derived from species distribution models for Lepidoptera in Great Britain
Cite this dataset as:
Rolph, S; Mondain-Monval, T.O.; August, T.; Jarvis, S.G.; Wright, E.; Fox, R.; Pocock, M.J.O. (2023). Species richness and recording priority derived from species distribution models for Lepidoptera in Great Britain. NERC EDS Environmental Information Data Centre. https://doi.org/10.5285/445381ce-f412-48a0-bc3c-2d0ef4737274
Download/Access
PLEASE NOTE: By accessing or using this dataset, you agree to the terms of the relevant licence agreement(s). You will ensure that this dataset is cited in any publication that describes research in which the data have been used.
This dataset is available under the terms of the Open Government Licence
https://doi.org/10.5285/445381ce-f412-48a0-bc3c-2d0ef4737274
This dataset includes a set of modelled outputs produced as part of the DECIDE project. Three groups were modelled; butterflies, day-flying moths and night-flying moths. (For the moths, we only considered 'macro-moths'.) For each group there are three outputs; species richness, model variability and DECIDE recording priority. The outputs summarise across multiple species within each group. The model’s prediction probability of occurrence for individual species is not made available. The outputs are in a raster format on Ordnance Survey National Grid reference system (OSGB) at 100m x 100m resolution.
Species richness layers are a modelled prediction of how many species are present at a location. Model variability is used to determine where a model is uncertain about its prediction of species occurrence. Model variability is combined with information about how recently a species had been recorded to produce the DECIDE recording priority. The DECIDE recording priority is a measure to prioritise locations to support adaptive sampling of where to collect species occurrence data to improve species distribution models.
Species richness layers are a modelled prediction of how many species are present at a location. Model variability is used to determine where a model is uncertain about its prediction of species occurrence. Model variability is combined with information about how recently a species had been recorded to produce the DECIDE recording priority. The DECIDE recording priority is a measure to prioritise locations to support adaptive sampling of where to collect species occurrence data to improve species distribution models.
Publication date: 2023-07-07
View numbers valid from 07 July 2023 Download numbers valid from 20 June 2024 (information prior to this was not collected)
Format
TIFF
Spatial information
Study area
Spatial representation type
Raster
Spatial reference system
OSGB 1936 / British National Grid
Provenance & quality
Outputs are derived from species distribution models (SDMs) fitted to species occurrence data and environmental data. The species occurrence data is from structured surveys and opportunistic species record. Data is from Butterflies for the New Millennium (https://butterfly-conservation.org/our-work/recording-and-monitoring/butterflies-for-the-new-millennium) and the UK Butterfly Monitoring Scheme and the National Moth Recording Scheme. Environmental data used for fitting SDMs included; BIOCLIM variables derived from long term average climate data, altitude, slope, aspect, land-cover (UKCEH land-cover map).
SDMs were fitted for species separately using an ensemble modelling approach using four different model types; logistic regression (GLM), general additive model (GAM), random forest (RF) and Maxent (ME). Best performing models for each species were used. Species richness is the summed predicted probability of presence across all species within each group (butterfly, day-flying moths, and night-flying moths).
Model variability was calculated using a bootstrapping approach. For each model type, we fit the model 10 times on 90% random subsets of the total species' occurrence dataset. Model variability was calculated as the standard deviation across each model’s predicted probability of occurrence at each location for each species. The model variability was summed across all species within each group to produce the model variability layer.
The DECIDE recording priority (where a recorder should visit next) was calculated as a composite of model variability and how recently a record had been made in a location. Specifically, the model variability was down-weighted by the days since each record by dividing by the number of months since last record.
In the DECIDE tool (https://decide.ceh.ac.uk), the DECIDE recording priority layer is updated daily as new records are made on iRecord, iNaturalist and iSpot. The data presented here is a snapshot of the layers as of 20-Feb-2023.
SDMs were fitted for species separately using an ensemble modelling approach using four different model types; logistic regression (GLM), general additive model (GAM), random forest (RF) and Maxent (ME). Best performing models for each species were used. Species richness is the summed predicted probability of presence across all species within each group (butterfly, day-flying moths, and night-flying moths).
Model variability was calculated using a bootstrapping approach. For each model type, we fit the model 10 times on 90% random subsets of the total species' occurrence dataset. Model variability was calculated as the standard deviation across each model’s predicted probability of occurrence at each location for each species. The model variability was summed across all species within each group to produce the model variability layer.
The DECIDE recording priority (where a recorder should visit next) was calculated as a composite of model variability and how recently a record had been made in a location. Specifically, the model variability was down-weighted by the days since each record by dividing by the number of months since last record.
In the DECIDE tool (https://decide.ceh.ac.uk), the DECIDE recording priority layer is updated daily as new records are made on iRecord, iNaturalist and iSpot. The data presented here is a snapshot of the layers as of 20-Feb-2023.
Licensing and constraints
This dataset is available under the terms of the Open Government Licence
Cite this dataset as:
Rolph, S; Mondain-Monval, T.O.; August, T.; Jarvis, S.G.; Wright, E.; Fox, R.; Pocock, M.J.O. (2023). Species richness and recording priority derived from species distribution models for Lepidoptera in Great Britain. NERC EDS Environmental Information Data Centre. https://doi.org/10.5285/445381ce-f412-48a0-bc3c-2d0ef4737274
Supplemental information
Correspondence/contact details
Dr. Simon Rolph
UK Centre for Ecology & Hydrology
Maclean Building, Benson Lane, Crowmarsh Gifford
Wallingford
Oxfordshire
OX10 8BB
UNITED KINGDOM
enquiries@ceh.ac.uk
Wallingford
Oxfordshire
OX10 8BB
UNITED KINGDOM
Authors
Other contacts
Rights holder
UK Centre for Ecology & Hydrology
Custodian
NERC EDS Environmental Information Data Centre
info@eidc.ac.uk
Publisher
NERC EDS Environmental Information Data Centre
info@eidc.ac.uk
Additional metadata
Keywords
Funding
Natural Environment Research Council Award: NE/V003054/1
Last updated
29 February 2024 16:54